Epigenetic Segregation of Microbial Genomes from Complex Samples Using Restriction Endonucleases HpaII and McrB.

Guohong Liu,Dan Sphar,Christopher Q Weston,Paula King,Shannon Waltz,Robert T Yamamoto,R Allyn Forsyth,Helen Barnes,Long K Pham,Gunnar F Kaufmann

doi:10.1371/journal.pone.0146064

Abstract

We describe continuing work to develop restriction endonucleases as tools to enrich targeted genomes of interest from diverse populations. Two approaches were developed in parallel to segregate genomic DNA based on cytosine methylation. First, the methyl-sensitive endonuclease HpaII was used to bind non-CG methylated DNA. Second, a truncated fragment of McrB was used to bind CpG methylated DNA. Enrichment levels of microbial genomes can exceed 100-fold with HpaII allowing improved genomic detection and coverage of otherwise trace microbial genomes from sputum. Additionally, we observe interesting enrichment results that correlate with the methylation states not only of bacteria, but of fungi, viruses, a protist and plants. The methods presented here offer promise for testing biological samples for pathogens and global analysis of population methylomes.

Highlights

Generation Sequencing (NGS) has expanded our perception of microbial diversity in the human microbiome [1] which plays roles in diverse clinical conditions such as obesity, allergies and cancer [2,3,4,5]
Since CpG methylation occurs frequently in eukaryotic genomes, we hypothesized that HpaII would bind and concentrate microbial genomic DNA, which have lower levels of m5C [16, 23], from mixtures containing human and higher eukaryotic genomic DNA
We examined enrichment profiles in genomic DNA mixtures

Summary

Introduction

Generation Sequencing (NGS) has expanded our perception of microbial diversity in the human microbiome [1] which plays roles in diverse clinical conditions such as obesity, allergies and cancer [2,3,4,5]. A key advantage of NGS in these studies is the non-hypothesis driven approach which allows detection of novel pathogens where primers or probes would have missed the causative agent [8, 9], as well as characterization of unexpected genes such as virulence factors in Staphylococcus aureus [10] and macrolide resistance in Mycobacterium tuberculosis [11]. For most clinical sample DNA preparations, microbes, pathogens, are typically present at trace levels resulting in inefficiently sequencing a vast majority of host DNA rather than the desired microbiome or causative pathogen. The epigenome of only a small number of bacterial species has been well defined [15,16,17], and epigenomes of protists, fungi and viruses remain poorly characterized

Methods

Results

Conclusion